TomasBM

That depends on the model and the toolkit it uses. In my experience from using Claude Code (Max, Opus 4.5+) intensely for the past six months, I maybe had 3 instances where the implementation broke functionally. And all of these breaking changes were resolved by Claude.

Obviously, this won't apply to every context: I work primarily with well-known langs (e.g., Python, JS), small to medium codebases (<500k LoC, for sure), and relatively few co-developers.

This is actually a good idea.

I mean, not sure if everyone wants that for their project, and there will surely be plenty of trade-offs.

But it would be a very good compromise: You (the maintainer) get only human-generated PRs in the canonical project, and they (pro-AI contributors) get a lower-threshold sandbox to play with. Best case scenario, you cherrypick the pre-filtered golden nuggets to bring back to the canonical project.

Isn't that what I'm already doing? I assume that the author believes that he's right often enough to argue his position(s), but feels dejected without being proven wrong at the end of the argument.

But there's another important point here: the answer to the "am I really right?" question isn't always clear at the start of every argument.

Unless you believe there's room for (dis)proving your position, or getting some nuance on a topic [1], it's not a debate or an argument - it's a lecture. And lectures depend on other social dynamics which don't apply here.

[1] For example, maybe there are other reasons behind the position that the person can't express easily, or maybe you're actually arguing about different things.

I like this - thanks for sharing!

Definite agree on having human-based redundant systems.

But I think the point the parent commenter made is: if there's not a functional difference in the result (i.e., job satisfies the definition of done), it doesn't matter if the AI generated the code or did the diagnosis.

But I think it's also fair to say that the process matters, even if the result is the same. If something exists for our benefit (e.g., there's no real alternative to learning-by-doing, and people need to know stuff for safety/security reasons), and we're fine with the trade-off, there's no reason to just give up the process to the AI.

Other than the obvious, self-reflective question that the author doesn't pose - "what if I'm the one who's wrong?" - I think it's worth arguing if the conditions are right.

Because I also like being correct, a debate to me has become something of a game where (ideally) we both win in both end scenarios: either my thinking was correct, and now I verified/validated it, and got you to think differently; or my thinking was incorrect, and you corrected it for me (or helped me get there).

However, I implicitly figured out that there are some qualifiers to actually getting the benefits:

- Can I be, and remain, polite and reflective? If not, my personality or knee-jerk responses will always get in the way of an argument's benefits.

- Is the subject sensitive to the person for whatever reason? If yes, any argument inadvertently becomes a signal of a person's worth.

- Are we in a competitive setting (e.g., corporate meeting, or larger social group)? If yes, any argument inadvertently becomes a social status competition.

- Do I know how to stick to the issue (instead of moving goalposts), and stop when the debate gets overwhelming (too long, too much difference)? If not, I'll overstep the boundary after which it isn't mutually beneficial anymore.

These are not easy to figure out, and sure, maybe stop arguing with most people if the conditions aren't right.

But unless you stop communicating altogether, I don't see how you can stop arguing with people in general.

Seems like this policy would apply pretty well regardless of who/what generated the code.

Well, unless you needed those million copies for whatever reason, that is an example of spam or denial-of-service, regardless of how it's generated.

And I'm not disagreeing - it is hard to anticipate what needs verifying, regardless if it's functional or non-functional.

But if it's not a spam submission, you could probably design tests or static/dynamic analysis tools that can verify those million copies much faster than manual reviews.

Good point. Even if you submit small, verifiable and readable changes, you can still overload the review process by submitting too many of them (e.g., 100s of PRs).

But I'd argue that some projects [1] could benefit from the speed (and sometimes, quality) of AI code generation without filtering by something that's difficult to identify (i.e., is it truly human-generated).

One way could be to constrain the size of each commit and PR, and invest more heavily into the review process (e.g., tests, static/dynamic analysis, sandbox deployments), so even if you get 100s of contributions, you can knock each out quickly.

Obviously, easier said than done. And at that point, you may as well use the AI to make the commits yourself, instead of relying on community contributions.

[1] Of course, this is only the case if the project's only purpose is to be a tool, and not also an educational reason for humans to learn how to code - in which case, it makes sense to invest more into identifying the "cheaters".

I think that's completely fair.

There are also plenty of valid personal reasons for refusing to generate code with AI - learning-by-doing and ownership of the result being the main ones, IMO.

everybody wants to let AI do their work for them, but nobody wants to be downstream of AI work.

This is also true in my experience. But in my work, I found that I don't care how the code or comment was generated, as long as it doesn't try to overload my brain with irrelevant and obfuscated things, and as long as the person is not pretending that it's true, verified or their own creation (when it isn't).

If you have some requirements/specifications, and the piece of code fits them, then it runs.

Alternatively, if you have some vague idea [1] about what you expect to see/have, and the running code satisfies that idea, then it also runs.

Obviously, there are plenty of non-functional specs (e.g., security, cleanness, readability) that a code should probably fulfill before one finds it acceptable, but these are also not somehow impossible for state-of-the-art models to satisfy.

[1] Vibe, if you prefer, tho I dislike the term. Another related term is eyeball estimation.

Current narrative where?

I don't see this narrative being more prominent than the "AI won't be displacing jobs because things worked out in the past" counternarrative.

For example, only recently has The Economist published articles that go more towards that narrative (AFAIK) [1].

there’s just not enough data to be conclusive.

With this, I can agree. We can still extrapolate based on what we currently know.

[1] https://www.economist.com/leaders/2026/05/14/prepare-for-an-...

Interesting initiative. What are the guiding reasons behind these lists?

I can't think of a functional reason for a no-AI policy: if it runs, it runs, regardless of who or what made it.

Also, even if you avoid AI-generated slop, you can't really avoid the human-generated or human+AI-generated slop that passes your filters.

Still, I can definitely think of good non-functional reasons: provenance, accountability, proof-of-work, encouraging people to write code themselves, empirically tracking how humans develop codebases, etc.

It's a fair policy. Getting those verbose, AI-authored walls of text is very annoying, especially when you're expected to thoroughly review it. It's like a denial-of-service attack on the human mind. I can only imagine how frustrating this can get in open projects that get a lot of contributions.

However, I don't think this will discourage AI-based coding at all. In fact, I see two potential outcomes of these policies:

- Negative: Submitters just add stylistic markers to make their accounts and output seem human-generated. This is like syntactic sugar: the core content and the size of contributions stay the same, but the style gets quirkier.

- Positive: Submitters actually provide to-the-point, no-bullshit commits and comments - "here's the code, here's why I made that change, here are the effects of that change". Even if AI-generated, these small contributions may become much easier to verify & validate. We may even see some standardization in terms of what qualifies as an appropriately sized contribution, what requires more thorough review (e.g., adding unverified dependencies), etc.

I personally wouldn't care if it was AI-generated or not, as long as the content fit the latter category.

Actually, the replication crisis shows how difficult (or underinvested) the process of reviewing is.

Removing this (often very basic) peer review doesn't somehow fix the problem. The solution lies in more thorough reviews and replication studies, not in everyone deciding for themselves.

I had even proposed and implemented an integration for arXiv Labs that got accepted, but then never merged. I should follow up on that...

You definitely should - looks like what I roughly had in mind.

Thanks for sharing!

ArXiv is a good complement to the modern peer review, IMO. As long as someone "vouches" for you, and you adhere to its minimal standards, you're able to post a paper. Other readers can decide whether the paper is worth their attention, and whether the presented ideas or results are valuable.

It's also good that it doesn't gatekeep with the paywalls that you can pretty much only afford by affiliating yourself with a toll-paying institution.

Obviously, there are plenty of flaws with this system:

1. If you're associated with a brand (e.g., Google, MIT) or have a recognizable co-author (e.g., Yann LeCun), you'll get attention and citations no matter what.

2. "Vouching" can also just mean accepting someone's email request without ever having met or known them.

3. It puts the effort on the readers to decide whether each paper is valuable, and particularly scientifically valuable, for which most readers will be unequipped.

4. "Minimal standards" can be gamed by AI-generated submissions.

I'd love to see a synthesis of arXiv, open-access publishing and artifact reviews, like the following:

- Have a number of reviewers on retainer, or design a reward system similar to bug bounties. The reward mechanism probably shouldn't be based on money or allow a winner-takes-all strategy.

- Have a number of badges with respect to the quality and value of the paper. For example: validated by peers (i.e., reviewed by at least 3 peers with minimum borderline accept consensus), valuable (i.e., reviewed by at least 5 peers with a valuable indicator), etc.

- Allow vouched comments on the platform, and moderate for self-promotion, toxicity, etc. Obviously a big ask.

- Improve the "vouching" system, or add badges like "vouched by X people" or "vouched by established scientist".

Hope their new organization will implement some of these improvements.

Unfortunately, I don't think systems thinking alone would help much.

One could present the case in favor of Internet age verification to the nth-order effects, while downplaying the effects in the case against.

So, in addition to presenting the cases with foreseeable effects, we need ways to compare the impact of worst-case scenarios in the two cases, and make a decision or compromise based on that.

Why is it psychosis and not lower standards?

While I can understand being skeptical of non-experts' claims that such answers are enough, I don't understand why you call it "psychosis" and not simply naivety or lack of expertise.

At the same time, the new so-called "models" haven't been pure transformer-based LLMs, but entire systems with tools (with access to the Internet), data storage, and the options to trigger additional instances for different tasks.

I'm not always a fan of the squiggles, but I can appreciate the UI pattern. It's definitely one of the more intuitive and recognizable visual markers for "something's wrong with this word".

Isn't that bias just legitimized by having culture fitness criteria in hiring?

Your scenario is confusingly formulated, but it ultimately deals with moral risk.

If you suspect that it's fraud, you should seriously consider whether you want to be a part of it, even if you can't prove the fraud. Examples are companies centered around cryptocurrencies: not all of them end up being rugpulls, but enough of them do, making every new company potentially morally risky for each engineer.

If you only suspect that it's an overly ambitious project, but not fraud, the moral risk is obviously lower. Sure, it may be economically risky - as all unproven business models are - or you may discover the fraud later (like the author apparently did). Examples are startups like SpaceX in its initial stages: there was nothing ostensibly fraudulent in trying to make and sell better rockets, just very economically risky.

Obviously, the devil is in the details, but if you're honest with yourself, you should do your due diligence (i.e., does it seem like fraud or not) and determine your moral risk appetite. It's up to the wider society to reduce and punish the unacceptable levels of moral risk, regardless of the appetites of individual engineers.

Do you have a specific example of a wasteful STEM research project that was cut?

My (perhaps wrong) impression was that wastefulness was given as the reason for making the cuts, but that the cuts were done broadly and indiscriminately [1].

In other words, the actions don't match the stated goal of reducing wastefulness. They seem more like a punishment for the members of all scientific institutions, and deterrence for curiosity-driven research.

[1] For example, the cuts to the STEM grants & projects didn't seem attached to any evidence of said projects' wastefulness.

No disagreement there.

I would even argue that anything that enables broad-area targetting since the invention of the Maxim machine gun and artillery would qualify as a weapon of mass destruction. In an ideal world, the first machine gun or dynamite would've made everyone see war as a very bad resolution mechanism since the 19th century.

But here we are, so the game is to not make war much worse than it already is.

You're interpreting the claims more strongly than they were presented, IMO.

One of its premises is that The Rich are some cohesive group

There doesn't have to be a fully cohesive group of an entire class for there to be negative consequences for the non-members of that class. The members also don't have to be cohesive or aligned at first, but they will tend to align on issues that threaten their position.

For example, we have historically seen that the incumbent elites tend to be anti-socialist/-communist, because their relative position and power are threatened, even if their populations and some (aspiring) elites are remotely pro-socialist. And because the coercive power of incumbents is often much higher than the power of some populace or aspirants (i.e., they have more to offer to those who hold coercive power), they will tend to succeed in pushing against the majority.

The entire history of the US, UK, Germany, France, Russia (etc.) is full of such examples.

that can trade amongst themselves in a hermetically sealed economy.

It doesn't have to be hermetically sealed to exclude certain populations.

On a global level, many countries don't trade with each other, and their economies are doing fine. Hard sanctions and cold wars even make this intentional, rather than a product of "we don't have anything to offer to each other".

On a local level, most people don't interact with the "homeless"/unhoused, and the latter often don't have much to offer to the former. Most Western countries don't need to hermetically separate the rich from the rest, but if you look at Brazil, South Africa, Saudi Arabia or even the US, gated communities are common in some areas. Most of the rest outside don't have much to offer to those inside.

Taken together: if it's plausible, and we don't know how probable it may be, we may want to figure that out, instead of hoping it's not probable.

Can confirm. From my experience, since December 2025, state-of-the-art models can replace a lot of need for junior, medior or senior developers.

This meant that, in my example and given the same scope, one junior person with AI could easily handle multiple projects in 2026 that would require 2 or more medior people as recently as 2025. The only reasons for involving multiple people were time, learning opportunity, or increased scope - not skill.

If we paused AI progress today, and ignored the option of adding agents recursively, the current state of things still provides several alternatives to any honest & informed manager:

- Altruistic or optimistic: Keep all current jobs, and move the scope until there's a need and/or funds to hire more devs;

- Pragmatic or neutral: Keep the current jobs, but not hire anyone else as long as the scope is balanced against time (our current situation);

- Cold or pessimistic: Lay off anyone who doesn't fit the leanest model at current (or acceptably lower/higher) scope.

I don't know whether this translates into a "job's crisis", and I can only hope it won't. But hope is not a career strategy, scope is not inexhaustible, and the political, economic and social pressures are quite strong even if you removed AI.

The difference is in "active"/intelligent versus "passive"/dumb targetting that's performed by the machine.

The missile, once fired, has the general vicinity (if not the exact position) of the target and is armed by the operator. Therefore, the operator is fully accountable for the targetting. Same goes for the landmines, once placed. Hitting civilians is reckless at best, and negligent at worst.

An autonomous weapon system (AWS) usually means that the system, once deployed, can do the targetting itself over any arbitrarily bounded location. An AWS can continue finding targets as long as its hardware allows it. For kamikaze drones, it's one time; for other drones, the ammunition & battery are the limits.

We currently rely on human targetting because we assume that A) humans are able to distinguish between legitimate and illegitimate targets "well-enough", and if not, B) at least we can hold them accountable (e.g., punish them for war crimes).

An AWS provides a layer of plausible deniability: the operator can claim that the system wasn't developed well enough, while the developer can claim that it wasn't used as intended. Given the inscrutability of modern computational intelligence - i.e., visual-action neural networks - this could potentially lead to very worrying incidents.

From a technical POV, the difference between a manually operated drone and an AWS drone may not be massive. From a military POV, it's just another legal lethal tool in the arsenal.

But from a social/civilian POV, the use of AWS is still 'not normal' and opens a can of worms. Targetting while evading counterattacks and crimes successfully is a bottleneck for manual operation. That's no longer the case with AWS: build 20 thousand drones, for example, and you can trivially win by overwhelming any manual defense of frontlines or cities. And knowing the history of human warfare, winning can range from relatively bloodless regime changes to utter destruction of the loser's civilization.

So, the best outcome is similar to nuclear deterrence or MAD: as long as everyone has 20 thousand AWS drones, they're safe.

Look, I'm not even claiming that piracy is a good term for it, or that the association between theft and copying wasn't pushed in pro-IP campaigns. But for the past 20 (or 30?) years, piracy has been the word everyone uses for this type of activity.

My point is: even if you changed the word, and explained the difference to everyone who couldn't distinguish between (A) literal theft of a physical object, and (B) a replication of some sequences of binaries, you're still not addressing the real question.

And the real question is: Why shouldn't an author have an option to protect their IP, physically or digitally? What's the difference between copying a physical book and a digital book without author's approval, from the point-of-view of an author trying to get paid for every copy?

I'm personally a fan of alternatives to IP-based earnings and earning enforcement, but I don't see how that helps any author trying to make a living from their content in the world where there are very few alternatives.

It's not necessarily incompatible: authors can make money in ways that don't depend on enforcing IP or even the number of books sold. For example, Patreon, Kickstarter, government subsidies, payment for number of books written, grants, etc.

However, all those other ways are more difficult to set up, and can be risky for the funders, so IP enforcement is the least-worst solution.

While I also find various approaches and reasons behind IP governance dumb, copying IP is piracy. In practice, copying some things is unlawful, regardless if we think it shouldn't be or if piracy once only referred to naval burglary.

If you copy a book in a bookstore, or leave a perfect synthetic copy of a natural diamond you take, you'll likely be charged with something. Digitally, that's much a clearer legal charge because copying is easier. So, neither is theft, but that doesn't make it lawful either.

There are valid reasons for enforcing IP rights digitally, because "all content should always be free" doesn't pay the bills when all you (can or want to) do is produce content. No existing society agrees that content producers should be subsidized, so in a society dependent on "earn for yourself", content producers shouldn't be punished.

But the punishment does exceed the severity of the "crime" by a lot, I agree.